JP7373273B2 - Image processing device, image processing method, medical information processing device, medical information processing method, radiography system and program - Google Patents

Description

The present invention relates to an image processing device and an image processing method that process medical images, a medical information processing device that generates medical information, a radiation imaging system, and a program.

Medical information processing devices have functions such as providing information based on user trends and preferences using machine learning, and improving image analysis accuracy. Patent Document 1 describes a method of improving image recognition accuracy by machine learning and detecting a target object. Further, Patent Document 2 describes a method of recognizing a division pattern of a radiation image, an irradiation field, an imaging position, and an imaging region using a neural network.

Japanese Patent Application Publication No. 2017-185007 Japanese Patent Application Publication No. 4-261649

When configuring the medical information processing apparatus to perform inference processing by machine learning, it is necessary to input learning result data obtained in advance by machine learning to the inference processing unit. When the amount of learning result data obtained through machine learning increases, there is a process of loading the learning result data from a storage medium such as a hard disk, and a process of expanding the read learning result data to memory in order to make it usable for inference processing. It will take time. Furthermore, a large amount of unnecessary learning result data is included in the inference processing, which may lead to deterioration of the quality of the processing. As a result, it is no longer possible to quickly display appropriate medical images, which reduces the operability of the medical information processing device.

An object of the present invention is to realize rapid processing by selectively acquiring learning result data of machine learning according to test information.

An image processing apparatus according to one aspect of the present invention for achieving the above object has the following configuration. That is,
A plurality of learning result data obtained in advance by machine learning, including learning result data regarding irradiation field recognition and learning result data regarding detection of multiple different types of object parts or lesion parts, Learning in which one set includes the learning result data regarding the irradiation field recognition and at least one learning result data selected from the learning result data regarding the detection of the plurality of different types of object parts or lesion parts based on the examination information. reading means for reading the result data group ;
A process of determining an irradiation field in the medical image by performing inference processing on the medical image acquired based on the examination information using the read learning result data group , and a process of determining the subject part in the medical image. or processing means for performing a process corresponding to the read learning result data group among a plurality of processes including a process for determining a lesion part.

According to the present invention, rapid processing can be realized by selectively acquiring learning result data of machine learning according to test information.

FIG. 1 is a diagram illustrating a configuration example of a radiation imaging system according to a first embodiment. 5 is a flowchart showing photographing processing according to the first embodiment. 5 is a flowchart showing learning result data acquisition processing according to the first embodiment. (a) is a diagram showing an example of learning result data acquisition information, and (b) is a diagram showing an example of test information. The figure which shows the example of a structure of the radiation imaging system by 2nd Embodiment. 12 is a flowchart showing learning result data acquisition processing according to the third embodiment.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. Note that in the following embodiments, the term radiation may include, for example, α rays, β rays, γ ray particle rays, cosmic rays, etc. in addition to X-rays. Further, the following embodiments can be combined as appropriate, and forms in which the embodiments are appropriately combined are also included in the embodiments of the present invention.

<First embodiment>
FIG. 1 is a diagram showing an example of the configuration of a radiation imaging system according to the first embodiment. The radiography system includes a control device 100, a radiation detection device 110, an operation section 120, a radiology information system, a radiology information system (RIS) 130, a display section 140, and a radiation generation device 150. The control device 100 is connected to the RIS 130 and controls radiography using the radiation detection device 110 and the radiation generation device 150.

The radiation detection device 110 detects radiation emitted from the radiation generation device 150 and passed through a subject (not shown), and outputs image data corresponding to the radiation. Note that the image data can also be referred to as a medical image or a radiation image. Specifically, the radiation detection device 110 detects the radiation that has passed through the subject as an electric charge corresponding to the amount of transmitted radiation. For example, the radiation detection device 110 includes a direct conversion type sensor such as a-Se that converts radiation into electric charges, a scintillator such as CsI that converts radiation into visible light, and a-Si sensor that converts radiation into visible light. An indirect sensor using a photoelectric conversion element such as the following is used. Further, the radiation detection device 110 generates image data by A/D converting the detected charges, and outputs the image data to the control device 100.

The control device 100 is connected to the radiation detection device 110 via, for example, a wired or wireless network or a dedicated line. The radiation detection device 110 images the radiation generated by the radiation generation device 150 and outputs image data to the control device 100. The control device 100 has an application function that runs on a computer. That is, the control device 100 has one or more processors and a memory, and the processor executes a program stored in the memory to implement each functional unit described below. However, part or all of each functional unit may be realized by dedicated hardware. The control device 100 performs image processing on the image data output from the radiation detection device 110, generates an image, and displays the image on the display unit 140. The operation unit 120 receives instructions from an operator. Further, the control device 100 has a function of controlling each component. The control device 100 outputs an image to the display unit 140 and provides a graphical user interface using the display unit 140 while controlling the operation of the radiation detection device 110.

The control device 100 controls the timing at which the radiation generating device 150 generates radiation and the radiation imaging conditions. In the control device 100, the image acquisition unit 101 controls the timing at which the radiation detection device 110 captures and outputs image data. The examination information input unit 104 is an example of a first acquisition unit that acquires examination information. The examination information input unit 104 of this embodiment inputs examination information manually input by the operator from the operation unit 120, or acquires examination information from the RIS 130 and allows the user to select it using the operation unit 120. The examination information input to the examination information input section 104 is managed in association with image data photographed by the radiation detection apparatus 110 based on the examination information.

The learning result data acquisition unit 105 reads learning result data from the learning result data storage unit 106. The learning result data storage unit 106 stores learning result data obtained by machine learning using teacher images. Further, the learning result data acquisition unit 105 includes information that associates various words and word combinations included in the examination information with the learning result data stored in the learning result data storage unit 106. Therefore, the learning result data acquisition unit 105 can acquire learning result data used in the processing of the image processing unit 102 (inference processing unit 103) based on the words included in the examination information. The learning result data acquisition unit 105 is an example of a reading unit that reads learning result data selected based on the test information from a storage unit (learning result data storage unit 106) that stores learning result data acquired in advance by machine learning. It is.

The image acquisition unit 101 is an example of a second acquisition unit that acquires a medical image acquired based on the examination information acquired by the first acquisition unit (examination information input unit 104). In this embodiment, a radiation image photographed by the radiation detection apparatus 110 is acquired as a medical image. The image processing unit 102 performs image processing such as noise removal on the image data output from the radiation detection device 110. The image processing unit 102 can also perform image processing such as trimming and rotation on the image output from the radiation detection device 110. The inference processing unit 103 performs inference processing using learning result data by machine learning, such as irradiation field recognition and gradation processing. The image processing unit 102 may include a plurality of inference processing units as the inference processing unit 103 depending on purposes such as irradiation field recognition and gradation processing. The image processing unit 102 causes the display unit 140 to display the image after image processing. The image processing unit 102 and the inference processing unit 103 are examples of processing units that process acquired medical images using learning result data read out by a reading unit (learning result data acquisition unit 105).

Next, radiographic image processing according to the first embodiment will be described according to the flowchart in FIG. 2. In step S201, the examination information input unit 104 acquires a plurality of pieces of examination information from the RIS 130. In step S202, the learning result data acquisition unit 105 determines learning result data to be read out for each of the plurality of pieces of inspection information acquired in step S201, and acquires learning result data indicating the correspondence between the inspection information and the learning result data to be read out. Generate information. The learning result data acquisition information will be described later with reference to FIG. 4(a). In step S203, the test information input unit 104 allows the user to select one of the plurality of test information acquired from the RIS 130 and sets it as the test target. Such processing may include, for example, displaying a plurality of pieces of acquired test information in a list format, and setting the selected test information as a test target in response to an operation input by a user who selects one piece of test information from the list. It is realized by Note that the user may directly input the test information from the operation unit 120. In that case, the selection operation in step S203 may be unnecessary.

In step S204, the control device 100 starts the inspection by transmitting a signal for transitioning the radiation detection device 110 to the preparation state according to the set inspection information. In response to this signal, for example, the radiation detection device 110 controls the bias power supply using the main control circuit, and applies a bias voltage to the two-dimensional image sensor. After that, in order to read out the dark current signal accumulated in the pixel, the drive circuit performs initialization to read out the image signal from the pixel array. After the initialization is completed, the radiation detection device 110 transmits status information to the control device 100 indicating that it is ready to obtain a radiation image. Further, the control device 100 (examination information input unit 104) sets operating parameters (tube voltage, etc.) of the radiation generating device 150 based on the examination information selected in step S203. When the control device 100 receives the status information from the radiation detection device 110 indicating that preparations for imaging are complete, the control device 100 notifies the radiation generation device 150 of permission for exposure.

In step S205, the image acquisition unit 101 acquires a radiation image taken by the radiation detection device 110. More specifically, for example, when the radiation generating device 150 that has been notified of permission for exposure irradiates radiation in response to the operation of the exposure button, the drive circuit of the radiation detecting device 110 detects the irradiated radiation and detects the radiation obtained. The readout circuit reads out the image signal and generates a radiation image. The radiation detection device 110 transmits the generated radiation image to the control device 100. The image acquisition unit 101 of the control device 100 receives this radiation image. In this way, the examination information input unit 104 and the image acquisition unit 101 of the control device 100 control the operation of radiography based on the examination information, and the radiographic images obtained by the radiography are taken based on the examination information. It functions as a control unit for acquiring radiographic images.

On the other hand, in step S206, the learning result data acquisition unit 105 selectively selects learning results from the learning result data storage unit 106 based on the examination information selected in step S203 and the learning result data acquisition information generated in step S202. Get data. Details of the process in step S206 will be described later with reference to the flowchart in FIG. Here, the examination information includes, for example, order information, subject information, examination type, etc. shown in FIG. 4(b). For example, the order information includes the area and direction of radiography, the subject information includes the attributes of the subject (e.g., body thickness, elongation, weight, gender, age), and the type of examination indicates whether the examination is a regular examination or an emergency examination. Includes the type of test. Note that the process in step S205 may be executed in parallel with the process in step S204 (obtaining a radiation image).

In step S207, the image processing unit 102 performs diagnostic image processing on the radiation image acquired by the image acquisition unit 101. At this time, the inference processing unit 103 executes inference processing by machine learning using the learning result data acquired in step S204. The inference processing unit 103 performs, for example, a process of analyzing the feature amount of an image, determining an irradiation field, a process of determining display contents such as annotations to be superimposed on the image, and the like. Further, in the diagnostic image processing, the results of such processing may be displayed on the display unit 140. In step S207, the control device 100 ends the test in response to the input operation by the operator.

Next, the learning result data acquisition process (the process of step S205) by the learning result data acquisition unit 105 will be described with reference to the flowchart in FIG. 3. In step S301, the learning result data acquisition unit 105 acquires the examination information selected (confirmed) in step S203. In step S302, the learning result data acquisition unit 105 refers to the learning result data acquisition information generated and stored in advance in step S202, and selects learning result data according to the test information acquired in step S301. Then, in step S303, the learning result data acquisition unit 105 reads the learning result data selected in step S302 from the learning result data storage unit 106, expands it so that it can be used by the inference processing unit 103, and stores it (not shown). keep it in the section.

The learning result data acquisition information has, for example, the content shown in FIG. 4(a). According to the learning result data in FIG. 4(a), in the case of test information [1], learning result data A, B, and E are acquired, and in the case of test information [2], learning result data A, C , D, and E are shown. Furthermore, in the case of test information [3], it is shown that learning result data A and C are acquired. The learning result data acquisition unit 105 refers to such learning result data acquisition information and selectively acquires necessary learning result data according to the examination information. For example, learning result data A is related to the irradiation field, learning result data B is related to the anatomical features of the chest, learning result data C is related to the anatomical features of the abdomen, and examination information [5] is the chest frontal image. Assume that there is a shooting order for . In this case, it is not necessary to acquire at least the learning result data C related to the abdomen, and in the example of FIG. 4(a), the learning result data acquisition information is configured so that the learning result data A and B are acquired. . Therefore, when test information [5] is acquired in step S301, the learning result data acquisition unit 105 reads the learning result data A and B from the learning result data storage unit 106 (steps S302 and S303).

As described above, according to the first embodiment, since the learning result data corresponding to the examination information is selectively acquired from the learning result data storage unit 106, the time required to acquire the necessary learning result data is shortened, and the The processing of the radiation image by the processing unit 102 can be executed quickly. Further, the image processing unit 102 can implement processing (for example, inference processing) using appropriate learning result data. As a result, a desired medical image can be obtained quickly, and the operability of the medical imaging apparatus is improved.

<Second embodiment>
In the first embodiment, the configuration (FIG. 1) in which the control device 100 includes the learning result data storage section 106 has been described, but the present invention is not limited to this. The learning result data storage unit 106 may be provided in an external storage device communicably connected to the control device 100. FIG. 5 is a block diagram showing a configuration example of a radiographic system according to the second embodiment. In FIG. 5, the learning result data storage section 106 is arranged in an external storage device 200 provided outside the control device 100. In addition, in FIG. 5, the same reference numerals are given to the same components as in the first embodiment. Examples of the external storage device 200 include a network storage, an external computer, a cloud, and the like.

In order for the control device 100 to acquire learning result data, it is necessary to communicate between the control device 100 and the external storage device 200, but this communication may be in any form, such as wired or wireless. Good too.

According to the second embodiment as described above, similarly to the first embodiment, since learning result data corresponding to examination information is selectively acquired, it is possible to improve the operability of the medical imaging apparatus. . Further, in the second embodiment, since the learning result data can be shared with the control device 100 of another facility, there is an advantage that the latest learning result data can be easily distributed and managed.

<Third embodiment>
In the first embodiment, the learning result data acquisition information is referred to, learning result data corresponding to the acquired examination information is selected, and the selected learning data is read from the learning result data storage unit 106. In the third embodiment, if the learning data selected based on the examination information has already been acquired, control is performed so as not to read it. In this way, by avoiding reading out duplicate learning data, processing can be further reduced. The configuration of the radiographic system according to the third embodiment is the same as that of the first embodiment (FIG. 1). The operation of the learning result data acquisition unit 105 according to the third embodiment will be described below using the flowchart shown in FIG. Note that the process shown in FIG. 6 replaces the process described in FIG. 3.

Similar to the first embodiment, the learning result data acquisition unit 105 acquires the examination information selected in step S203 (step S701). Then, the learning result data acquisition unit 105 selects learning result data to be acquired from the acquired examination information and the learning result data acquisition information generated in step S202 (step S702).

The learning result data acquisition unit 105 has a holding unit that holds the read learning result data. In steps S703 to S705, the learning result data acquisition unit 105 reads out, from the learning result data storage unit 106, learning result data that is not held in the holding unit among the learning result data selected based on the examination information in step S702. Ru. First, in step S703, the learning result data acquisition unit 105 determines whether one of the learning result data determined to be acquired in step S302 is already acquired learning result data. If it is determined in step S703 that the learning result data is unobtained learning result data, the learning result data acquisition unit 105 reads the learning result data from the learning result data storage unit 106 in step S704. On the other hand, if it is determined in step S703 that the learning result data has already been acquired, step S704 is skipped, and new learning result data is not read. In step S705, the learning result data acquisition unit 105 determines whether the above processing has been performed on all of the learning result data selected in step S702, and if there is unprocessed learning data, the process returns to step S703. . If it is determined in step S705 that all of the selected learning data has been processed, the process shown in FIG. 6 ends.

Note that, in the above description, whether or not to read the learning result data is determined based on whether or not the learning data to be read is held in the holding unit; however, the present invention is not limited to this. For example, learning result data other than the learning result data held in the holding unit may be specified by referring to the learning result data acquisition information, and the specified learning result data may be additionally read and held in the holding unit. good.

As described above, according to the third embodiment, not only learning result data corresponding to examination information is selectively acquired, but also learning result data that has not been acquired in consecutively executed examinations is acquired. As a result, learning result data can be obtained more quickly. Further, in a configuration in which the read learning result data is developed and held in a form that can be used by the inference processing unit 103, it is also possible to avoid executing duplicate development processing.

<Fourth embodiment>
In the third embodiment, unobtained learning result data is additionally obtained and retained. Therefore, if there is a large amount of test information, there is a possibility that the storage capacity of the storage unit of the learning result data acquisition unit 105 will become insufficient. In the fourth embodiment, for learning result data groups that are in an exclusive relationship, one of them is always held, thereby reducing the occurrence of storage capacity strain in the holding unit. That is, when reading the learning result data selected based on the inspection information, the learning result data acquisition unit 105 discards from the holding unit the learning result data that has an exclusive relationship with the newly read learning result data. Operate. The operation of the learning result data acquisition unit 105 according to the fourth embodiment will be described below. Note that the configuration of the radiation imaging system is the same as that of the first embodiment (FIG. 1).

In the control device 100, the operation of the learning result data acquisition unit 105 will be explained using the learning result data acquisition information shown in FIG. 4(a) and the flowchart shown in FIG. In the example of FIG. 4(a), learning result data B and C are data used for inference processing for the same purpose, and are used exclusively. The learning result data acquisition information stores related information indicating that the learning result data B and C have an exclusive relationship. The learning result data acquisition unit 105 acquires and discards learning result data according to the examination information and its related information. That is, in step S704, the learning result data acquisition unit 105 acquires learning result data, and if learning result data having an exclusive relationship with the acquired learning result data is held, it discards it.

For example, in FIG. 4(a), when tests are performed in the order of test information [1], [2], and [3], test information [1] acquires learning result data A, B, and E and makes inferences. The information is input to the processing unit 103. In the next test information [2], since the learning result data A and E have already been obtained, the learning result data C and D are obtained and input to the inference processing unit 103. At this time, the retained learning result data B is discarded because it has an exclusive relationship with the acquired learning result data C. In the next test information [3], the learning result data A and C have already been input to the inference processing unit 103, so the learning result data acquisition process can be omitted. Note that the combination of learning result data in an exclusive relationship may be determined from the learning result data acquisition information generated in step S202, or may be set in advance.

As described above, according to the fourth embodiment, regarding the acquisition process of learning result data used in inference processing for the same purpose, learning result data is discarded based on the exclusive relationship of the learning result data. Therefore, it is possible to reduce the strain on the storage capacity in the storage unit of the learning result data acquisition unit 105. Note that the learning result data that is required regardless of the test information may be made mandatory, and the other learning result data may be treated as optional. In that case, the optional learning result data may be arbitrarily selected according to the quality and accuracy of the radiation image requested by the user. Alternatively, usage fees for optional learning result data may be charged separately.

<Fifth embodiment>
In the first embodiment, learning result data corresponding to the test information is acquired through the processes of S301 to S303 and used for inference processing. In the fifth embodiment, the learning result data acquisition unit 105 of the control device 100 specifies learning result data (learning result data selected for all test information) that is used independently of test information. The learning result data is read and retained regardless of the selection of inspection information. More specifically, the identified learning result data is read out and acquired in advance from the learning result data storage unit 106 prior to selecting the examination information.

For example, in the learning result data acquisition information in FIG. 4(a), learning result data A is selected for all test information. For example, learning result data A is the learning result regarding the irradiation field, and is the learning result data that is commonly used in the inference processing of irradiation field recognition that is commonly performed on radiation images acquired for all examination information. . The learning result data acquisition unit 105 acquires such common learning result data in advance, for example, prior to the selection of examination information in step S203. Then, in steps S301 to S303 in FIG. 3, the learning result data acquisition unit 105 acquires learning result data other than the learning result data acquired in advance, according to the selected examination information.

As described above, according to the fifth embodiment, learning result data that does not depend on examination information can be acquired before the examination starts, so the learning result data to be acquired at the time of imaging can be reduced, and the required It is possible to obtain learning data.

According to each of the embodiments described above, it is possible to selectively acquire learning result data of machine learning used for inference processing according to examination information, and to quickly display an appropriate medical image. As a result, the operability of the medical imaging device can be improved. In addition, even when the amount of learning result data used by the machine learning inference processing unit increases, the learning result data can be read and processed by selectively acquiring the learning result data corresponding to the inspection information from a storage device such as a hard disk. This makes it possible to make the expansion process to memory more efficient.

Note that in each of the above embodiments, an example has been shown in which image processing such as irradiation field recognition and gradation processing is performed on a medical image using learning result data, but the present invention is not limited to this. For example, a medical information processing device may be provided that generates medical information for a medical image using acquired learning result data. Here, the medical information is, for example, diagnostic support information such as segmentation in a medical image or a lesion detected in a medical image. In this case, the medical information processing device acquires the examination information and reads learning result data selected based on the examination information from a storage unit that stores at least one or more learning result data acquired in advance by machine learning. The medical information processing device then acquires the acquired medical image based on the acquired examination information, and generates medical information for the acquired medical image using the learning result data read out from the storage unit. do. According to such a configuration, by selectively acquiring and using learning result data of machine learning according to examination information, it becomes possible to quickly display medical information for a medical image.

<Other embodiments>
The present invention provides a system or device with a program that implements one or more of the functions of the embodiments described above via a network or a storage medium, and one or more processors in the computer of the system or device reads and executes the program. This can also be achieved by processing. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

100: Control device, 101: Image acquisition unit, 102: Image processing unit, 103: Inference processing unit, 104: Examination information input unit, 105: Learning result data acquisition unit, 106: Learning result data storage unit, 110: Radiation detection 120: Operation unit, 130: RIS, 140: Display unit

Claims (16)

A plurality of learning result data obtained in advance by machine learning, including learning result data regarding irradiation field recognition and learning result data regarding detection of multiple different types of object parts or lesion parts, Learning in which one set includes the learning result data regarding the irradiation field recognition and at least one learning result data selected from the learning result data regarding the detection of the plurality of different types of object parts or lesion parts based on the examination information. reading means for reading the result data group ;
A process of determining an irradiation field in the medical image by performing inference processing on the medical image acquired based on the examination information using the read learning result data group , and a process of determining the subject part in the medical image. or processing means for performing a process corresponding to the read learning result data group among a plurality of processes including a process of determining a lesion part. 2. The apparatus according to claim 1, further comprising an acquisition unit that controls a radiographic operation based on the examination information and acquires a radiographic image obtained by the radiography as a medical image photographed based on the examination information. image processing device. The examination information includes at least one of the following information: order information including the radiation imaging region and imaging direction, object information regarding the attributes of the object, and information indicating the examination type including the types of normal examination and emergency examination. Item 2. The image processing device according to item 1 or 2. The image processing apparatus according to any one of claims 1 to 3, wherein the reading means reads the learning result data group from a storage means provided inside or outside the image processing apparatus. When the first test information among the plural test information is selected,
The reading means reads out a learning result data group including one set of learning result data regarding the irradiation field recognition and learning result data regarding the detection of the first subject part selected based on the first examination information. ,
The processing means performs a process of determining an irradiation field in a medical image acquired based on the first examination information and a process of determining the first subject part,
When the second test information among the plurality of test information is selected,
The reading means reads out a learning result data group including one set of learning result data regarding the irradiation field recognition and learning result data regarding the detection of a second subject part selected based on the second inspection information. ,
5. The processing means performs a process of determining an irradiation field in a medical image acquired based on the second examination information and a process of determining the second subject part. The image processing device described in . Further comprising a selection means for allowing the user to select one of the plurality of inspection information,
2. The readout means reads out a learning result data group including one set of learning result data regarding the irradiation field recognition and learning result data selected based on one piece of examination information selected by the selection means. 6. The image processing device according to any one of 5 to 5. Further comprising a generating means for determining learning result data to be read out for each of the plurality of pieces of test information and generating learning result data acquisition information indicating a correspondence between the test information and the learning result data to be read out,
7. The image processing apparatus according to claim 6, wherein the reading unit refers to the learning result data acquisition information and reads the learning result data corresponding to the selected piece of examination information. Further comprising a holding means for holding the learning result data read by the reading means,
2. The reading means reads learning result data selected based on the test information that is not held in the holding means from the storage means storing the plurality of learning result data. 7. The image processing device according to any one of 7. Further comprising a holding means for holding the learning result data read by the reading means,
8. The reading means specifies learning result data other than the learning result data held in the holding means by referring to the learning result data acquisition information, and additionally reads out the identified learning result data. Image processing device. 10. According to claim 8 or 9 , the reading means discards learning result data having an exclusive relationship with newly read learning result data from the holding means when reading the learning result data selected based on the test information. The image processing device described. According to claim 7 or 9 , the reading means specifies learning result data to be used independently of test information from the learning result data acquisition information, and reads the identified learning result data regardless of selection of test information. The image processing device described. 12. The image processing apparatus according to claim 11 , wherein the reading means reads out the identified learning result data prior to selecting examination information. The image processing apparatus according to any one of claims 1 to 12 , wherein the reading unit reads learning result data selected based on words included in the examination information. An image processing device according to any one of claims 1 to 13 ;
a radiation detection device that captures the medical image;
A radiography system equipped with A plurality of learning result data obtained in advance by machine learning, including learning result data regarding irradiation field recognition and learning result data regarding detection of multiple different types of object parts or lesion parts, Learning in which one set includes the learning result data regarding the irradiation field recognition and at least one learning result data selected from the learning result data regarding the detection of the plurality of different types of object parts or lesion parts based on the examination information. a reading step of reading the result data;
A process of determining an irradiation field in the medical image by performing inference processing on the medical image acquired based on the examination information using the read learning result data group , and a process of determining the subject part in the medical image. or a processing step of performing a process corresponding to the read learning result data group among a plurality of processes including a process of determining a lesion part. A program for causing a computer to function as each means of the image processing apparatus according to any one of claims 1 to 13 .

Images